Colorectal cancer is the third most common cause of cancer death. Surgery, in conjunction with adjuvant chemotherapy and radiation therapy, is the most common and effective treatment for colorectal cancer. However, even in patients with presumed resectable cancer, the threat of unrecognized occult disease remains a major challenge for the surgeon. Therefore, 40% of the patients ultimately develop recurrent or metastatic disease. In patients with distant metastases, long-term survival rates fall to below 10%.To date, no reliable systems have been developed to provide the surgeon with real-time intraoperative information to accurately predict the adequacy of resection (both surgical margins and occult disease). More than half of the colorectal cancer patients with recurrent disease, however, cannot undergo surgical procedure due to the tumor's critical location. For these patients with recurrent colorectal cancer, the standard chemotherapy yields a 15% to 50% response rate, so new therapeutic regimens need to be explored. Considering that most cancers generally have multiple genetic abnormalities, a single drug targeting a particular oncogene is unlikely to be completely effective. In this regard, geldanamycin (GA) and its analogs provide a hope for treating the advanced colorectal cancer by inhibiting the molecular chaperone Hsp90 to simultaneously down-regulate many oncogenes. Their severe toxicity, though, has limited their clinical evaluation, and new modifications or drug delivery methods are required for future clinical application. Therefore, we intend to develop a system for intraoperative real-time detection of occult tumors and abnormal lymph nodes, as well as to provide a therapeutic regimen for unresectable and advanced chemo- resistant colorectal cancer. The long term goal is to integrate intraoperative tumor detection for surgery and targeted drug therapy into one system with the same tumor targeting antibody. We hypothesize that the anti-TAG-72 antibody (HuCC49ACH2)-glycosidase conjugate will target gross tumors, occult tumors, and lymph nodes involved in the disease process, and will be used in radioimmunoguided surgery (RIGS) to accurately assess the resection margin and ultimately improve patient survival. In cases where surgery is impossible, we will take advantage of the tumor-localized antibody-enzyme conjugate for site-specific activation of the geldanamycin prodrug in the tumor. We hypothesize that the site-specific activation of GA prodrug will increase the local active drug concentration in the tumor improving its anticancer efficacy, while the prodrug will remain inactive in normal tissue (due to the lack of antibody- enzyme conjugate) and thereby reducing its side effects. Three specific aims are proposed in the project: Specific aim I: To deliver the anti-TAG-72 antibody-galactosidase conjugate (AbE) to colorectal cancer for tumor detection in surgery and for future prodrug activation. Specific aim II: To synthesize geldanamycin-galactose prodrugs (GA-prodrug) with enzyme specific activation properties in tumors. Specific aim III: To site-specifically activate the geldanamycin prodrug (GA prodrug) by the tumor-localized antibody-enzyme (AbE) in order to increase their efficacy and decrease their side effects. In summary, this integrated system offers a strong clinical application to revolutionize the surgical procedure and provides a new targeted drug therapy for advanced chemo-resistant colorectal cancer.